All of the money from subscriptions and advertisements is put into a big pool (minus a cut for Spotify), and every month it is divided up between each rightsholder based on the proportion of plays. So if you have a big chunk of track plays that are just generated noise playing over and over for hours, that's a big chunk of that pool going to unoriginal/easily reproducible uploads instead of actual musicians. It's basically a scam gaming the way the system works.
It's also costly for Spotify. Even if the streams were in the form of, say, podcasts that were not allowed any sort of monetization, it's still hours upon hours (per user) of data that has to be streamed...and it doesn't compress efficiently. Compression algorithms seek to avoid noise, and deliberately generated noise will not compress well. So the amount of data being streamed is much higher per second than with music or speech.
Meanwhile, you can make an app that plays white/pink/whatever noise trivially. No waste of resources necessary.
And English speakers are only a fraction of the user base. Current events in the US social media bubble barely penetrate the general public in the US, let alone across international and language barriers.
It's probably the largest social media platform in Japan, for example.